SAKURA: On the Multi-hop Reasoning of Large Audio-Language Models Based on Speech and Audio Information

Open in new window